Re: [sage-devel] weird commits on top of develop

2023-02-27 Thread John Cremona
We could also have a branch called release which is separate from develop. Normal developers would make PRs to be merged into develop like now, but the release manager could make PRs from develop to release. That's the way we do things in the LMFDB (with different names for the branches). So deve

Re: [sage-devel] weird commits on top of develop

2023-02-27 Thread Tobias Diez
The branch protection rules don't allow to make an exception for a single person (the release manager). What can be configured though is that all pushes to develop (from everybody, including admins) have to go through a PR. This would mean that the merge and release commits also has to go throu

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Matthias Koeppe
Branch protection is on, but currently admins can bypass it. This can be tightened in the config here if so desired: https://github.com/sagemath/sage/settings/branch_protection_rules/33965567 On Sunday, February 26, 2023 at 12:36:42 PM UTC-8 Volker Braun wrote: > I agree that we should turn on

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Dima Pasechnik
On Sun, Feb 26, 2023 at 4:29 PM Vincent Delecroix <20100.delecr...@gmail.com> wrote: > > In other words, everytime you got confused we will have a develop > branch not aligned with the most recent tag and a bunch of extra > commits cancelling each other? I don't think I really can constructively r

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Volker Braun
I agree that we should turn on branch protection, it is rather easy to accidentally push to develop. Its also not ok to push what you think is a CI fix without it going to review, we had some irregularities to support the github transition but that is over now. On Sunday, February 26, 2023

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Vincent Delecroix
In other words, everytime you got confused we will have a develop branch not aligned with the most recent tag and a bunch of extra commits cancelling each other? On Sun, 26 Feb 2023 at 16:46, Dima Pasechnik wrote: > > sometimes Volker misses things which make CI break, and such a release is > pr

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Dima Pasechnik
sometimes Volker misses things which make CI break, and such a release is pretty much useless, in the sense that you can't gauge anything from the CI results on PRs made on top of it. Or an important CI or other infrastructure improvement doesn't make it into the release. In previous release cycle

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Vincent Delecroix
I do not understand why develop isn't protected against anybody but the release manager pushes. On Sun, 26 Feb 2023 at 11:16, Dima Pasechnik wrote: > > it was a mistake on my side - I thought develop branch was broken, but it was > not. > > On Sun, 26 Feb 2023, 10:13 Vincent Delecroix, <20100.de

Re: [sage-devel] weird commits on top of develop

2023-02-26 Thread Dima Pasechnik
it was a mistake on my side - I thought develop branch was broken, but it was not. On Sun, 26 Feb 2023, 10:13 Vincent Delecroix, <20100.delecr...@gmail.com> wrote: > Hello, > > Why do we have commits on top of develop after the release commits? Namely > > commit 52a81cbd161ef4d5895325657c88a68590

[sage-devel] weird commits on top of develop

2023-02-26 Thread Vincent Delecroix
Hello, Why do we have commits on top of develop after the release commits? Namely commit 52a81cbd161ef4d5895325657c88a68590ea1d3b Author: Dima Pasechnik Date: Fri Feb 24 21:15:42 2023 + Revert "add missing # optional - gap3" This reverts commit c017a6a3d68f0aca8cb63ceba52fe451a64